Unconventional Shale Gas-Condensate Reservoir Performance: Impact of Rock, Fluid, and Rock-Fluid Properties and their Variations

Transcription

1 SPE MS/URTeC: Unconventional Shale Gas-Condensate Reservoir Performance: Impact of Rock, Fluid, and Rock-Fluid Properties and their Variations A. Orangi and N. R. Nagarajan, Hess Corporation, Houston, Texas Copyright 2015, Unconventional Resources Technology Conference This paper was prepared for presentation at the Unconventional Resources Technology Conference held in San Antonio, Texas, USA, July The URTeC Technical Program Committee accepted this presentation on the basis of information contained in an abstract submitted by the author(s). The contents of this paper have not been reviewed by URTeC and URTeC does not warrant the accuracy, reliability, or timeliness of any information herein. All information is the responsibility of, and, is subject to corrections by the author(s). Any person or entity that relies on any information obtained from this paper does so at their own risk. The information herein does not necessarily reflect any position of URTeC. Any reproduction, distribution, or storage of any part of this paper without the written consent of URTeC is prohibited. Abstract Liquid-rich shale (LRS) reservoirs, particularly lean and rich gas-condensates, are economically attractive but pose unique production challenges. These include productivity losses caused by condensate banking, saturation pressure changes due to pore confinement, and the associated rock-fluid interactions. In recent years, development of these types of reservoirs has advanced with optimal lateral well placement and innovative completion designs. It has been well recognized that initial production rates, their decline during depletion, and the ultimate liquid recoveries are severely impacted by the fluid, rock, rock-fluid properties, and the completions design. Thus, it is critical to have an in-depth understanding of the controlling factors related to fluids, rock and rock-fluid parameters that affect the long-term production performance of these reservoirs. We have conducted detailed reservoir simulation studies based on a sector model to investigate the sensitivity of rock, fluid, and rock-fluid parameters, and modeling fracture properties on reservoir performance. The reservoir model consists of a horizontal well and a network of hydraulic, induced. and natural fractures embedded in the rock matrix. The model also includes the effects of changes in fluid saturation pressures due to pore confinement and altered fluid phase behavior in nano-pore environments, interfacial tension (IFT)-dependent gas-condensate relative permeability curves and end-points, and the matrix and fracture properties. Sensitivity runs allow comparisons of reservoir performance predictions of rates, condensate dropout and the resulting condensate blocking, and ultimate liquid recovery potential. We developed a number of gas-condensate fluid models covering a range of Condensate to Gas Ratios (CGRs) (50 to 250 stb/mmscf) and implementing dewpoint suppression/elevation in nano-pore matrix and its impact on liquid dropout and condensate banking. In this article, we illustrate the results of the simulation study using a relatively rich condensate with a CGR ~100 stb/mmscf. We also used interfacialtension dependent relative permeability curves and critical condensate saturations (S cc ) for the matrix and for fractures. The compaction effects of matrix and the fractures during depletion were also accounted for. We conducted a number of sensitivity runs, including elevation/suppression of fluid saturation pressures, IFT- dependent relative permeability, critical condensate saturations, and condensate blocking. Based on the results of this study, the following conclusions can be drawn:

2 2 SPE MS/URTeC: Well rate and its decline with time (pressure) are highly sensitive to the fluid, rock, and the rock-fluid parameters, such as the dewpoint pressure, the relative permeability Corey exponents and ends point saturations. Ultimate liquid recovery is impacted by the condensate dropout characteristics, IFT, and saturation pressure changes. The liquid recovery is impacted by the leaning of produced fluids with pressure decline, mobilization of dropped-out condensates, and IFT effects on the relative permeability. Condensate blocking severely affects gas rates and hence the well productivity. The results served as a basis for and guided reservoir development strategies and production planning. Introduction Production from liquid-rich shale (LRS) reservoirs has gained significant momentum over the past decade because of their huge resource potential and economic impact. Optimal development and efficient management of these highly complex resources require new technologies and a step change in the existing production methodologies. A first step in this process is to develop a better understanding of these unconventional rocks and fluids contained in them and their impacts on production performance. Unconventional shale formations often constitute both the source and the reservoir and often contain a wide range of hydrocarbon fluid types based on source rock maturity, from dry gas to rich gascondensates and volatile oils. Over the past decade, significant progress in understanding geological, geochemical, geo-mechanical, petro-physical characteristics of these reservoirs has been made (Clarkson et al. [2011]) and improved characterization of the reservoir rocks, fluids, and rock-fluid interactions (Orangi [2011], Honarpour, [2012], Nagarajan, [2013], Bertoncello [2013]) have been accomplished. This paper investigates the sensitivity of rock and fluid parameters and rock-fluid interactions on gascondensates reservoir performance and the ultimate recovery potential. LRS reservoirs have the following general characteristics (Clarkson et al. [2011]): Hydrocarbon source rock (organic-rich matter), finely inter-bedded in the matrix with natural fracture networks providing flow conduits Heterogeneity and anisotropy in rock and rock-fluid properties (amounts of total organic content, TOC [115%]) Nano scale pore throat size (1-20 nm) Low matrix permeability ( nano-darcies) Varying wettability (oil wet to mixed wet) characteristics The variability in fluid pressure/volume/temperature (PVT) properties (API gravity [30-55], gas/oil ratio (GOR) [ scf/stb]) Changes in micro-fracture (MF) and matrix (M) permeability under high stress conditions during depletion Multi-fractured lateral wells are critical for LRS reservoirs to connect the network of milli/micro-darcy fractures and nano-darcy matrix. Proper placement of laterals and optimal stimulation techniques are two major factors to maximize reservoir contact and thus maximize production. One of the main challenges in predicting performance of these unconventional reservoirs is to obtain high quality rock and fluid data and assess their uncertainties to improve rock, fluid, and rock-fluid characterization for reliable reservoir simulation results and enhance stimulation modeling. The main focus of this paper is to develop improved characterization of, and then assess the impacts of properties of fluids and rocks, and rock-fluid interactions on performance prediction.

3 SPE MS/URTeC: Utica Example The Utica shale spreads over a large portion of the North Eastern United States and part of Canada in the Appalachian Basin and consists of formations of middle Ordovician age. The Utica formation in eastern Ohio and western Pennsylvania, in particular, is a mix of black calcareous organic-rich shale consisting of inter-bedded limestone and is sandwiched between the upper Utica shale and the lower Trenton or Lexington limestone. The formation thickness changes between 25 and 200 ft at depths of 4000 to ft (Murphy, et al [2013]). The hydrocarbon fluids in the eastern Ohio Utica formation range from dry gas to lean and rich gas-condensates and volatile oils farther in the Northwest. The reservoir temperature varies between 150 F and 200 F. The reservoir pressures gradients are generally high and range from 0.8 psi/ft to 0.55 psi/ft from East to West, with the highly over-pressured dry gas reservoirs located in the East while slightly over-pressured liquid rich formations in the West. The stock tank liquid gravity falls between 60 API (lean condensate/wet gas) and 45 API (rich condensate/volatile oil). Due to the large areal variations of fluid composition, the condensate yield, NGL content, and the gas heating values (BTU) of the produced fluids vary widely. Table 1 provides basic PVT properties of expected fluids from Utica shale reservoirs based on the coal ranking. Fig. 1a 1c provide the maps of fluid windows estimated from Trenton formation: a) thermal maturity measurements [after USGS], and b) theoretical overpressure, and (c) CGR from reflectance (Ro) values and CGR correlations.

4 4 SPE MS/URTeC: Table 1 Maturity cross reference with notional values of hydrocarbon fluid properties Figure 1a-c Maps of Fluid Window, Over Pressure and CGR

5 SPE MS/URTeC: Fluid Variations and Modeling The wide range of fluids and their varying properties need to be characterized, preferably with a single model based on an equation of state (EOS) compositional description. Reservoir fluid data from these unconventional plays are generally limited. Some of the data are from laboratory measurements while others come directly from field measurements. Often, the laboratory data may not be quite representative of in-situ fluids due to sampling challenges. In addition, field measurements, particularly the CGR, may not be consistent because of the conditions at which these measurement were obtained were not reported. Therefore, utmost caution should be exercised while screening the data for quality and using it in the fluid characterization. Thus, a combination of limited laboratory PVT properties and field-measured data were used in developing a series of fluid models to cover a wide range of fluids contained in these reservoirs. The initial model was developed for a lean condensate having a CGR of ~20 stb/mmscf. Basic laboratory data such as the fluid composition, constant composition expansion, and constant volume depletion data were available for this fluid. The conventional EOS model building procedure as outlined below was employed: Characterize C7 compositions using one of the industry standard procedures (e.g., Gamma distribution model). This procedure provides a reasonable number of pseudo-components (about 3 to 5) to describe the C 7 content of the fluid with their compositions and the estimated EOS properties (Tc, Pc, etc.). A PVT software package was used for characterizing the fluid. The input to the model was the pure components (~10), the pseudo-components and their properties, and a combination of laboratory and field data. An automatic regression procedure was used to match the data to create the initial model. The model was then fine-tuned by tweaking appropriate model parameters and the C 7 component distribution to get an improved match to the data. The original model was used as the base model to create additional models to capture a range of fluids with CGRs from 50 to 250 stb/mmscf. The fluid models with higher CGRs ( 200 stb/mmscf) exhibit volatile oil characteristics. The procedure below was followed to adopt the original fluid model for a range of fluids with the CGR range above: The lean CGR model was flashed through a two-stage separator and the resultant liquid and gas compositions were recombined to different GOR values to create several preliminary models with increasing CGRs. Approximately ten fluid models were created with CGRs ranging from 50 to 450 stb/mmscf. The models were then tuned to match a number of selected field data such as CGRs and stock tank oil API gravity, where available. The field data was screened with diligent QC procedures to select the most representative and consistent data sets. Finally, a systematic lumping procedure was employed to reduce the number of components to 9 and 6 components since the original 14-component model, when used in simulation, is time-consuming. The lumping procedure was further constrained to retain the data match of the original 14-component model. Thus, thirty fluid models were created overall, with three sets of models (14-, 9-, and 6-components) per CGR and for ten different CGRs. The component slates of these models are provided in Table 2.

6 6 SPE MS/URTeC: Table 2 Component slate of the three models per CGR Full Model Lumped Model 1 Lumped Model 2 N2 C1N2 C1N2 CO2 C2 C2C3CO2 C1 C3CO2 C4C5 C2 C4 C6-C14 C3 C5 C15-C19 i C4 C6-C14 C20 n C4 C15-C19 i C5 C20 nc5 C6 C7-10 C11-14 C15-19 C20 The compositional changes of the fluids are shown in the spider plot in Fig. 2. The C 1 contents changes from low 90 mole% to high 60 mole% while the C 7 content of the fluid varies over a range of 1 to 15 mole % for a wide range of CGRs. The NGL content of the fluids is between low 20 and low 30 mole %. In addition, Figs. 3a to 3d display the C 1 and C 7 compositional changes, NGL contents, condensate yields, and saturation pressure variations of these fluids as functions of CGR. The produced gas BTU ranges over 1000 BTU to 1300 BTU extracted from dry gas to near critical volatile oils. Figure 2 Compositions of Unconventional Reservoir Fluids with Varying CGRs

7 SPE MS/URTeC: Figure 3a-d The Variations of Compositions, NGL Content, Total Liquid Yield, and Saturations Pressures The pressure-temperature phase envelopes of these fluids are shown in Fig. 4. The phase envelopes cover the fluids ranging from dry gas to a fluid with a CGR of 200 stb/mmscf. Finally, the liquid dropout characteristics of these fluids in the contant composition expansion (CCE) and constant volume depletion tests are shown in Figs. 5 and 6. Figure 4 Predicted Phase Envelopes for Fluids with a Range of CGRs

8 8 SPE MS/URTeC: Figure 5 Liquid dropout curves in CCE test for Utica fluids Figure 6 Liquid Dropout Curves in CVD Test-Utica fluids Reservoir Rock Characterization Generally, the shale formation is a mixture of muddy sandstones and carbonates with varying contents of different minerals and total organic carbons (TOC). The formation consists of two major rock types: viz., quartz/calcite-rich (CR) and organic-rich (OR) rocks distributed throughout the reservoir. The rock characterization should address these different types of rocks with regard to routine and special core analysis for the best characterization of the formation. Core data acquisition starts from proper coring methods and core plug preparation and selection. This is followed by basic and special core analysis (SCAL) measurements. The basic core data includes porosity, permeability, fluid saturations, and pore/pore throat size distribution. SCAL data include wettability, capillary pressure, relative permeability, and electrical properties. Basic Core Measurements Measurements of this data on unconventional cores are highly challenging and are made more difficult because measurements can differ by orders of magnitude due to the nano-meter (nm) size of pores and pore throats. Direct application of conventional methods or tight rock or Coal Bed Methane (CBM)

9 SPE MS/URTeC: methods may not work well with these unconventional rocks. Basic property measurement techniques for shale samples are an extension of conventional methods employed in tight gas sands or CBM. They need further validation when applied to nano- Darcy permeability shale formations. Although Gas Research Institute (GRI) crushed sample analysis provides quick data, effective porosity, Klinkenberg-corrected air, the liquid permeability and fluid saturation data should be validated using data on core samples under net confining stress (NCS) as porosities on shale core plugs as low as % have been measured with a high repeatability ( / PU). Furthermore, it is noted that Boyles Law and Darcy flow seem to hold for these ultra-low pore throat sizes making core plug measurements more reliable. Special Core Analysis Special core analysis data such as wettability, capillary pressure, relative permeability, and electrical properties are very sparse because of formidable challenges for measurement. Conventional techniques for drainage and imbibition capillary pressure measurements are not applicable to ultra-low permeability shale due to experimental difficulties caused by high entry pressure required to initiate these measurements, resulting in severe stress condition inside the core plugs before the commencement of the measurement. Laboratory relative permeability measurement techniques on unconventional shale core plugs are not well established and suffer from several experimental challenges as given below: Presence of heterogeneities and discontinuities such as lamina or bitumen-filled pore space and fractures result in non-representative data. Two-phase flow may take place in two different pore networks. An alternative method to laboratory measurement is to employ imaging techniques coupled with the construction of a 3-D digital rock model and Lattice Boltzmann Model (LBM) flow simulation (Nagarajan, et al [2013]). The imaging technique employs a high-resolution 3-D FIB/SEM technology and argon ion milling technique wherein thin layers of several nm thick are milled with continuous imaging of these slices by FIB/SEM (Bertoncello [2013]). The data helps to construct a digital 3-D volume of rock. The LBM technique simulates multi-phase flow to calculate the relative permeability to flowing phases. This method of determining relative permeability curves has several drawbacks. The elementary rock volume is too small to be representative of flow units at core level, and obviously at reservoir scales. Core Data Used The basic core data for different rock types were obtained by conventional core analysis methods subjecting the data to stringent QA/QC protocols and diligent validation. Table 3 provides these basic core properties for the rock types, quartz/calcite rich (CR) and organic rich (OR) rocks. The porosities ( ) are in the range of 5 to 8.5 % with lower values associated with CR rocks. The absolute permeability (k a )of OR rocks is low while the CR rocks are more permeable. This is reflected directly in the effective permeability (k eff ) of these rock types. The rock compressibility of OR formation is twice as much as that of CR rocks. The initial water saturation (S wi ) of OR formation is lower, ~10%, while CR rocks have higher S wi of ~20%.

10 10 SPE MS/URTeC: Table 3 Basic rock properties for organic-rich and calcite-rich formations Basic Rock Property Organic Rich Quartz/Calcite Rich Porosity ( ) 6-8.5% 7.25% 5-7.5% 6.25% Absolute Permeability (K a ), nd Effective Permeability (K eff ), nd Compressibility, psi -1 12*10-6 6*10-6 Initilal Water Saturation SCAL Data Used A combination of core-level laboratory measurements (Dacy [2010]), coupled with the insight developed on the underlying physics of flow through nanometer pores by pore level modeling such as Lattice Boltzmann Method (Clague [2000], Wagner [2008]), helped to construct meaningful relative permeability curves for the shale gas- condensate fluids. In general, gas-condensate simulation studies assume constant relative permeability curves during the life of the reservoir. However, they can change significantly with pressure depletion, in particular for very rich gas condensates, as the gas-oil interfacial tension changes significantly from less than 0.1 dynes/cm (in the near- critical region) to as high as several dynes/cm as the fluid moves away from this region. The changing gas- condensate interfacial tension effects were included in the relative permeability curves developed. The key relative permeability (k r ) parameters such as critical condensate saturation (S cc ), trapped gas saturation (S gt ). and the Corey exponents determine the shape and the end points of the k r curves. When the first drop of condensate forms (at the dewpoint pressure), gas relative permeability (k rg ) experiences a sharp drop and subsequent condensate saturation build-up further reduces k rg, regardless of whether the condensate is mobile or not. Depending on the rock-fluid interactions and pore network geometry, the condensate flow begins only after the condensate saturation builds up to S cc. Fig. 7 provides the relative permeability curves used in this study for both the CR and OR matrix. These curves incorporate the impact of interfacial tension on the shape of the relative permeability curves. Straight lines passing through the origin with unit slope represent the fracture relative permeability curves. The OR matrix generally exhibits higher S cc as well as a higher condensate Corey exponent compared to the CR matrix. On the other hand, the OR rocks exhibit lower trapped gas saturation in contrast of the CR rock type. Figure 7 Typical Relative Permeability Curves for Calcite-Rich and Organic-Rich Rocks

11 SPE MS/URTeC: Simulation Study: 3D sector model A detailed 3-D sector model representing a horizontal well with several stage of hydraulic fractures, each stage consisting of four clusters and 45 natural and induced micro fractures distributed uniformly along and orthogonal to hydraulic fractures, was set up. The model served as a reference platform to test the impact of various characteristics of the rocks, fluids, and rock-fluid interactions discussed earlier. The sector model has the dimensions of 1100 ft long (x), 220 ft wide (y), and 150 ft thick (z), and is shown in Fig. 8. The model consists of 11 layers of alternating beds of OR and CR rock types. A series of equally distributed micro fractures were placed in these layers, perpendicular to the hydraulic fractures, to simulate natural and induced micro fracture networks. Fig. 9 illustrates the schematic of the hydraulic and micro fracture configuration. Figure 8 The Sector Model for Simulation Study Figure 9 Schematic of Hydraulic and Micro Fracture Configuration Logarithmic grid refinement around the hydraulic and micro fractures using the Tartan gridding method allowed accurate tracking of pressure and fluid saturation changes in the vicinity of the fractures. In this study, the fracture width selected was 0.5 ft, which adjusts the effective permeability in the fracture accordingly. Although it is desirable to keep the fracture cell width at lower values (e.g., 0.5 in.),

12 12 SPE MS/URTeC: significantly higher computational costs may forbid this approach, particularly when performing compositional simulation with very many components to capture retrograde liquid dropout characteristics of gas-condensate fluids accurately. Each hydraulic fracture was connected to 45 micro fractures with varying half-lengths of 25 ft close to wellbore and 2.2 ft at the tip of hydraulic fractures. Although it may be desirable to use different micro fracture half-lengths for the CR and OR rock types due to their differing characteristics, same halflength values were used for both the rock types to manage the computational costs. Matrix and Fracture Property Changes during Depletion Rock compaction during depletion significantly affects both the matrix and fracture flow properties as the net overburden stress strongly influences matrix and fracture permeability. In addition, the hydraulic fractures, generally categorized into propped and un-propped sections, respond differently to rock compaction. The effective permeability of the matrix, propped, and un-propped sections could degrade by orders of magnitude as the reservoir pressure depletes. The impacts of matrix and fracture compaction are captured in the simulation by using systematically varying pressure-dependent transmissibility multipliers (transmissibility functions) displayed in Fig. 10. As shown in Fig. 10, the matrix permeability degradation is the highest followed by the un-propped and propped hydraulic fracture permeability. Figure 10 Transmissibility Multiplier Changes With Reservoir pressure Results and Discussions The main objective of the simulation study is to explore the impacts of the retrograde fluid behavior in confined pore systems, rock property changes due to compaction, and rock-fluid interactions, in particular condensate blocking, IFT dependent relative permeability curves, and endpoints. In the following section we discuss the results and their implications. Phase Behavior in Confined Pore Space It has been recognized that in unconventional tight formations, the pore proximity of the in-situ fluids affects the observed bulk gas condensate dewpoint pressures (Tindy [1966], Sigmund [1973], Lee [1989], Udell [1982], Brusilovsky [1992], Ping [1996], DeHoff [2006], Didar [2013]) and retrograde phase behavior (Durmishian [1964], Shapiro [2000 and 2001], Campos [2009], Devegowda [2012]). The

13 SPE MS/URTeC: magnitude of dewpoint pressure changes depends on the strength of capillary forces and the rock-fluid interactions, surface forces. The method of calculating dewpoint pressure changes in a single pore of any given radius has been well established. However, when applied to a system of pores (distributed pore-sizes), the method has to make a few assumptions that render the results more approximate, particularly when an average pore radius is assumed for a wide pore size distribution (as in the reservoir rocks). On the other hand, recent pure component condensation experiments conducted in nano-size channels etched on silicon wafers indicate a probable elevation of condensation pressures (Yin [2014]). In the absence of reliable models or strong experimental evidences to quantify dewpoint pressure changes caused by nanometer pores, the bulk dewpoint pressure of a fluid with fixed composition was varied arbitrarily over a narrow range of /-200 psi, modifying the fluid model parameters. The resultant fluid models were used in the simulation to illustrate nanometer pore effects on the reservoir performance. Fig. 11 shows the simulation results when the dewpoint pressure is varied between 3060 to 3480 psia. As illustrated in Fig. 11, a higher dewpoint results in earlier condensate dropout causing condensate blocking of gas flow, steep decline in gas rate and productivity index. Thus, the well performance and the well productivity index are degraded earlier in the production. On the other hand, in the case of the fluid with the lowest dewpoint pressure, these impacts are delayed as expected. Figure 11 The Effect of Changing Dewpoints on Well Productivity Index Impact of Retrograde Phase Behavior and Relative Permeability In order to capture the impact of retrograde liquid dropout and relative permeability on produced fluids, we compared compositions of produced fluid as a function of pressure predicted by simulations and the produced fluid data from laboratory constant volume depletion (CVD) tests both above and below the fluid dewpoint pressure. In the CVD test, below the dewpoint pressure (P d ), the dropped-out liquids are not produced and part of the produced fluid is either the single-phase gas above the dewpoint pressure or the equilibrium (leaner) gas below the dewpoint pressure. However, in the simulation, part of the dropped-out condensate below P d is produced when liquid hydrocarbon saturation exceeds S cc and part of the dropped out liquid is mobilized depending on the condensate relative permeability values. To compare CVD results more realistically with simulation results below P d, a portion of the dropped-out liquid (above

14 14 SPE MS/URTeC: S cc ) in the CVD test was added to the produced fluid and its composition recalculated. The fraction of dropped-out liquid added to CVD-produced gas is estimated using the ratio of condensate to gas relative permeability (k rc /k rg ) ratios at any given pressure and the corresponding liquid saturation. Figs. 12 through 16 compare these modified CVD compositions with the simulation results. The agreement is excellent given the complexity of the processes involved in the reservoir simulation compared to the simple approach we used. Figure 12 Comparison of Produced Fluids-Methane Composition-CVD vs. Simulation Figure 13 Comparison of Produced Fluids-Ethane Composition-CVD vs. Simulation

15 SPE MS/URTeC: Figure 14 Comparison of Produced Fluids C3-CO2 Composition-CVD vs. Simulation Figure 15 Comparison of Produced Fluids- C4-C5 Composition-CVD vs. Simulation Figure 16 Comparison of Produced Fluids-C6 Composition-CVD vs. simulation

16 16 SPE MS/URTeC: As expected, the compositions of the produced fluids from simulation and unmodified CVD results match exactly until the dewpoint pressure is reached and the liquid saturation builds up to S cc.inthe unmodified CVD results, the compositions of produced C 1 continually increase while C 6 content decreases with pressure. However, the modified CVD results closely follow simulation data. This confirms the quality and reliability of both CVD and simulation results. Furthermore, the condensate re-vaporization contributes very little to the increase in the liquid content of the produced fluids at low pressures (see Fig. 17). Only part of the C 4 -C 5 components of the dropped out condensate is re-vaporized while C6 fractions do not show any observable increase (Figs. 15 and 16). Another point to note is that in these tight formations, the increase in liquid production due to condensate mobilization above S cc or revaporization at low pressures is minimal as the condensate relative permeability is low in these tight formations. Figure 17 CGR Changes with Reservoir Pressure Decline Impact of Condensate Blocking Next, the sensitivity of the predictions to condensate blocking and its impact on the gas of rate were explored through simulations by varying the critical condensate saturations over a wide range. Fig. 18 shows the impact of condensate blocking with varying S cc between 10% and 35%. Above the dewpoint pressure (P d ), the steady reservoir pressure depletion with a constant drawdown constraint contributes to a steady drop in the gas rate. The part of the curve labelled as Steady gas rate drop-fluid compressibility change in Fig. 18 shows a slow drop in the gas rate from the initial reservoir pressure to the fluid dewpoint pressure as the gas compressibility is changing.

17 SPE MS/URTeC: Figure 18 Effect of Varying Critical Condensate Saturation on the Gas Rate As the reservoir pressure reaches Pd, the gas rate drops much more steeply caused by condensate accumulation in the pore space, in particular in pore throats, choking the gas flow. The portion of the curve in Fig. 18, labelled as Condensate banking-s c S cc indicates that right below the dewpoint pressure of ~3250 psia, a steep decrease in the gas rate is observed. As the condensate-accumulation exceeds S cc, condensate starts flowing. This brings in a further step-decrease in the gas rate (portion of the curve marked Condensate mobilization ), as the liquid flow takes up part of the connected pore network. This effect is seen more pronounced at the higher S cc of 35% as expected. When steady-state liquid flow is established and as further liquid saturation build-up occurs, the gas rate remains fairly steady (portion of the curve marked Constant gas rate ), because the liquids are not further encroaching and takes up additional pore space. At lower reservoir pressures of ~1800 psia, the constant drawdown constraint is switched to a constant flowing bottom-hole-pressure (FBHP) constraint for operational considerations. As a result, the gas rate in this region exhibits continued decline in the gas rate as seen above the dewpoint pressure. Summary and Conclusions Shale gas-condensate reservoir performance was studied through a series of compositional reservoir simulations that incorporate hydraulic, induced and natural fracture networks. The impact of retrograde condensation, relative permeability effects, and altered fluid phase behavior due to nanometer pore confinement were estimated using a 3D sector reservoir model in the simulation studies. A retrograde gas-condensate of ~100 stb/mmscf CGR served as an intermediately rich gascondensate fluid in this study. Appropriate routine core and SCAL data for different rock types (calcite-rich and organic-rich) based on a combination of laboratory measurements and modeling were utilized in the simulations. In addition, rock compaction effects were included to capture matrix and fracture property changes during reservoir depletion. Several simulations were performed to study the sensitivity and cumulative impacts of altered dewpoint pressures and condensate dropout characteristics (of the in-situ fluids caused by nanometer pore confinement) on the horizontal well performance.

18 18 SPE MS/URTeC: The impact of gas and condensate relative permeability characteristics such as Corey exponents and endpoints on the reservoir performance were quantified through a series of simulation studies. The flowing conclusions were drawn based on simulation results: The retrograde condensate dropout and the resultant condensate blocking were observed through the predicted steep reduction in gas flow rates just below the fluid dewpoint pressure. The initial immobile condensate build-up contributes to a sharp decline in gas rates followed by another step change in the rate as liquids are mobilized above the critical condensate saturation. This indicates that the condensate accumulation and eventual flow takes up part of the connected pore network thus causing a sharp reduction in the gas rates. The pore confinement of in-situ fluids in nanometer pore systems influences the fluid dewpoint pressure and thus the onset conditions for a sharp reduction in gas rates. Irrespective of the dewpoint pressure, the condensate banking effects are evident and seem to cause severe loss of well productivity. However, as expected, elevated dewpoint results in earlier rate decline and accelerates its negative impact on well productivity. The CGR changes resulting from the changes in C 6 content of the produced fluids are consistent with the expected rock-fluid interaction parameters such as the relative permeability exponents and the endpoint saturations. The agreement of the produced fluid compositions between the simulation predictions and the modified CVD calculations is excellent. Condensate re-vaporization at low pressures does not contribute much to the increase in the CGR of produced fluids at lower pressures. In shale reservoirs, the condensate recovery seems to be more severely impaired below the dewpoint pressure than in conventional reservoirs. The liquid recoveries do not increase noticeably below the dewpoint in these tight formations. Liquid recoveries estimated through a simple CVD test in the laboratory closely follows the simulation results. The reservoir operational constraints may force to switch well constraint from a constant drawdown to a constant flowing-bottom-hole-pressure near the end of the reservoir life. This further accelerates gas rate decline causing severe well productivity loss. Acknowledgements The authors would like to acknowledge the support and the approval by Hess Management to publish this paper. We would like to thank Jon Wallace with Hess Corporation and Matt Honarpour, formerly with Hess and currently with BHP Billiton, for several of their contributions to this paper. Thanks are also due to Jacob Rosenzweig for many useful discussions. References Bertoncello, A. and Honarpour, M. M., Standards for Characterization of Rock Properties in Unconventional Reservoirs - Fluid Flow Mechanism, Quality Control, and Uncertainties, SPE , presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, September, Brusilovsky, A. I., Mathematical Simulation of Phase Behavior of Natural Multicomponent Systems at High Pressures With an Equation of State, SPE 20180, SPERE, , February, Campos, M. D., Akkutlu, I. Y., and Sigal, R. F., A Molecular Dynamics Study on Natural Gas Solubility Enhancement in Water Confined to Small Pores, SPE , Presented at the SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, October, Clague, D. S., Kandhai, B. D., Zhang, R., Sloot, P. M. A., Hydraulic permeability of (un)bounded fibrous media using the Lattice Boltzmann method, Phys. Rev. E, 61, 616, 2000.

19 SPE MS/URTeC: Clarkson, C. R., Pedersen, P. K., Production Analysis of Western Canadian Unconventional Light Oil Plays, CSUG/ SPE , presented at the Canadian Unconventional Resources Conference held in Calgary, Canada, November, Dacy, J. Core Tests for Relative Permeability of Unconventional Gas Reservoirs, SPE MS, presented at SPE Annual Technical Conference and Exhibition held in Florence Italy., September, DeHoff, R., Thermodynamics in Materials Science, Chapter 12, pp , CRC Taylor & Francis, New York, Second Edition, Devegowda, D., Sapmanee, K., Civan, F., and Sigal, R., Phase Behavior of Gas Condensates in Shales due to Pore Proximity Effects: Implications for Transport, Reserves and Well Productivity, SPE , presented at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, October Didar, B. R. and Akkutlu, I. Y., Pore-Size Dependence of Fluid Phase Behavior and Properties in Organic Rich Shale Reservoirs, SPE , presented at the SPE International Symposium on Oil Field Chemistry, The Woodlands, Texas, April, Durmishian, A. G., Mamedov, Y. H., and Mirzadjanzade, A. J., Experimental Investigation of Hydrodynamic and Thermodynamic Properties of Gas Condensate Systems During Flow Through Porous Media, Investia Academii Nauk, USSR, No. 1, , Firincioglu, T., Ozkan, E., and Ozgen, C., Thermodynamics of Multiphase Flow in Unconventional Liquids-Rich Reservoirs, SPE , SPE Annual Technical Conference and Exhibition, San Antonio, October, Honarpour, M. M., Nagarajan, N. R., Orangi, A., Arasteh, F., and Yao, Z., Characterization of Critical Fluid, Rock, and Rock-Fluid Properties-Impact on Reservoir Performance of Liquid-rich Shales, SPE , Presented at the SPE Annual Technical Conference and Exhibition, SanAntonio, October, Lee, S. T., Capillary-Gravity Equilibria for Hydrocarbon Fluids in Porous Media, SPE 19650, Presented at the SPE Annual Technical Conference and Exhibition, San Antonio, TX, October, N. R. Nagarajan, M. M. Honarpour, F. Arasteh, Critical Role of Rock and Fluid - Impact on Reservoir Performance on Unconventional Shale Reservoirs, URTeC , Presented at the Unconventional Resources Technology Conference, Denver, CO, August, Orangi, A., Nagarajan, N., R., Honarpour, M. M., and Rosenzweig, J., Unconventional Shale Oil and Gas- Condensate Reservoir Production, Impact of Rock, Fluid, and Hydraulic Fractures, SPE , Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, The Woodlands, TX, USA, January, Ping G., and Liangtian, S., A Theoretical Study of the Effect of Porous Media on the Dew Point Pressure of a Gas Condensate, SPE 35644, presented at the SPE Gas Technology Symposium and Exhibition, Calgary, Alberta, Canada, May Shapiro, A. A., Potsch, K., Kristensen, J. G., and Stenby, E. H., Effect of Low Permeable Porous Media on Behavior of Gas Condensates SPE 65182, presented at the SPE European Petroleum Conference, Paris, France, October, Shapiro, A. A., and Stenby, E. H., Thermodynamics of multicomponent vapor-liquid equilibrium under capillary pressure difference, Fluid Phase Equilibria, 178, 17, Sigmund, P. M., Dranchuk, P. M., Morrow, N. R., and Purvis, R. A., Retrograde Condensation in Porous Media, SPEJ, , April Tindy, R., and Raynal, M., Are Test-Cell Saturation Pressures Accurate Enough?, Oil and Gas J, 64, 126, 1966.

20 20 SPE MS/URTeC: Udell, K. S., The Thermodynamics of Evaporation and Condensation in Porous Media, SPE 10779, presented at the California Regional Meeting of SPE, San Francisco, March Wagner, A. J., A Practical Introduction to the Lattice Boltzmann Method, A Report by Alexander J. Wagner, Department of Physics, North Dakota State University, Fargo, March Yin, X., Private Communication, Petroleum Engineering Department, Colorado School of Mines, Golden, Colorado.